RE: [PATCH v4 02/15] iommu: Report domain nesting info
From: Liu, Yi L
Date: Mon Jul 06 2020 - 09:17:17 EST
> From: Auger Eric <eric.auger@xxxxxxxxxx>
> Sent: Monday, July 6, 2020 9:01 PM
>
> On 7/6/20 2:20 PM, Liu, Yi L wrote:
> > Hi Eric,
> >
> >> From: Auger Eric <eric.auger@xxxxxxxxxx>
> >> Sent: Monday, July 6, 2020 5:34 PM
> >>
> >> On 7/4/20 1:26 PM, Liu Yi L wrote:
> >>> IOMMUs that support nesting translation needs report the capability
> >>> info
> >> need to report
> >>> to userspace, e.g. the format of first level/stage paging structures.
> >>>
> >>> This patch reports nesting info by DOMAIN_ATTR_NESTING. Caller can
> >>> get nesting info after setting DOMAIN_ATTR_NESTING.
> >>>
> >>> Cc: Kevin Tian <kevin.tian@xxxxxxxxx>
> >>> CC: Jacob Pan <jacob.jun.pan@xxxxxxxxxxxxxxx>
> >>> Cc: Alex Williamson <alex.williamson@xxxxxxxxxx>
> >>> Cc: Eric Auger <eric.auger@xxxxxxxxxx>
> >>> Cc: Jean-Philippe Brucker <jean-philippe@xxxxxxxxxx>
> >>> Cc: Joerg Roedel <joro@xxxxxxxxxx>
> >>> Cc: Lu Baolu <baolu.lu@xxxxxxxxxxxxxxx>
> >>> Signed-off-by: Liu Yi L <yi.l.liu@xxxxxxxxx>
> >>> Signed-off-by: Jacob Pan <jacob.jun.pan@xxxxxxxxxxxxxxx>
> >>> ---
> >>> v3 -> v4:
> >>> *) split the SMMU driver changes to be a separate patch
> >>> *) move the @addr_width and @pasid_bits from vendor specific
> >>> part to generic part.
> >>> *) tweak the description for the @features field of struct
> >>> iommu_nesting_info.
> >>> *) add description on the @data[] field of struct iommu_nesting_info
> >>>
> >>> v2 -> v3:
> >>> *) remvoe cap/ecap_mask in iommu_nesting_info.
> >>> *) reuse DOMAIN_ATTR_NESTING to get nesting info.
> >>> *) return an empty iommu_nesting_info for SMMU drivers per Jean'
> >>> suggestion.
> >>> ---
> >>> include/uapi/linux/iommu.h | 78
> >> ++++++++++++++++++++++++++++++++++++++++++++++
> >>> 1 file changed, 78 insertions(+)
> >>>
> >>> diff --git a/include/uapi/linux/iommu.h b/include/uapi/linux/iommu.h
> >>> index 1afc661..1bfc032 100644
> >>> --- a/include/uapi/linux/iommu.h
> >>> +++ b/include/uapi/linux/iommu.h
> >>> @@ -332,4 +332,82 @@ struct iommu_gpasid_bind_data {
> >>> } vendor;
> >>> };
> >>>
> >>> +/*
> >>> + * struct iommu_nesting_info - Information for nesting-capable IOMMU.
> >>> + * user space should check it before using
> >>> + * nesting capability.
> >> alignment?
> >
> > oh, yes, will do it.
> >
> >>> + *
> >>> + * @size: size of the whole structure
> >>> + * @format: PASID table entry format, the same definition with
> >>> + * @format of struct iommu_gpasid_bind_data.
> >> the same definition as struct iommu_gpasid_bind_data @format?
> >
> > right. yours is much better.
> >
> >>> + * @features: supported nesting features.
> >>> + * @flags: currently reserved for future extension.
> >>> + * @addr_width: The output addr width of first level/stage translation
> >>> + * @pasid_bits: Maximum supported PASID bits, 0 represents no PASID
> >>> + * support.
> >>> + * @data: vendor specific cap info. data[] structure type can be deduced
> >>> + * from @format field.
> >>> + *
> >>> + *
> >>
> +===============+===================================================
> >> ===+
> >>> + * | feature | Notes |
> >>> + *
> >>
> +===============+===================================================
> >> ===+
> >>> + * | SYSWIDE_PASID | PASIDs are managed in system-wide, instead of per |
> >>> + * | | device. When a device is assigned to userspace or |
> >>> + * | | VM, proper uAPI (userspace driver framework uAPI, |
> >>> + * | | e.g. VFIO) must be used to allocate/free PASIDs for |
> >>> + * | | the assigned device. |
> >>> + * +---------------+------------------------------------------------------+
> >>> + * | BIND_PGTBL | The owner of the first level/stage page table must |
> >>> + * | | explicitly bind the page table to associated PASID |
> >>> + * | | (either the one specified in bind request or the |
> >>> + * | | default PASID of iommu domain), through userspace |
> >>> + * | | driver framework uAPI (e.g. VFIO_IOMMU_NESTING_OP). |
> >>> + * +---------------+------------------------------------------------------+
> >>> + * | CACHE_INVLD | The owner of the first level/stage page table must |
> >>> + * | | explicitly invalidate the IOMMU cache through uAPI |
> >>> + * | | provided by userspace driver framework (e.g. VFIO) |
> >>> + * | | according to vendor-specific requirement when |
> >>> + * | | changing the page table. |
> >>> + *
> >>> + +---------------+-------------------------------------------------
> >>> + -----+
> >> Do you foresee cases where BIND_PGTBL and CACHE_INVLD shouldn't be
> >> exposed as features?
> >
> > sorry, I didn't quite get it. could you explain a little bit more. :-)
> For SYSWIDE_PASID I understand SMMU won't advertise it. But do you foresee any
> nested implementation not requesting the owner of the tables to bind and invalidate
> caches. So I understand those 2 features would always be supported?
I think BIND_PGTBL is optional as ARM will bind guest pasid table to host. As for
CACHE_INVLD, per current nesting implementations, looks all stage-1/level-1
owner should issue cache invalidation when stage-1/level-1 changed. But still
added it here per the comments from Kevin. :-)
"So far this assumption is correct but it may not be true when thinking forward.
For example, a vendor might find a way to allow the owner of 1st-level page
table to directly invalidate cache w/o going through host IOMMU driver. From
this angle I feel explicitly reporting this capability is more robust."
https://lore.kernel.org/linux-iommu/MWHPR11MB1645B09EBDC76514ADC897A68C6F0@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
Regards,
Yi Liu
> >
> >>> + *
> >>> + * @data[] types defined for @format:
> >>> + *
> >>
> +================================+==================================
> >> ===+
> >>> + * | @format | @data[] |
> >>> + *
> >>
> +================================+==================================
> >> ===+
> >>> + * | IOMMU_PASID_FORMAT_INTEL_VTD | struct iommu_nesting_info_vtd
> |
> >>> + *
> >>> ++--------------------------------+---------------------------------
> >>> +----+
> >>> + *
> >>> + */
> >>> +struct iommu_nesting_info {
> >>> + __u32 size;
> >>> + __u32 format;
> >>> + __u32 features;
> >>> +#define IOMMU_NESTING_FEAT_SYSWIDE_PASID (1 << 0)
> >>> +#define IOMMU_NESTING_FEAT_BIND_PGTBL (1 << 1)
> >>> +#define IOMMU_NESTING_FEAT_CACHE_INVLD (1 << 2)
> >> In other structs the values seem to be defined before the field
> >
> > not sure. :-) I mimics the below struct from uapi/vfio.h
> Yep I noticed that afterwards. In IOMMU uapi it looks the opposite though. So I
> would alignto the style in the same file but that's not a big deal.
I see, may align with other iommu uapi. :-)
Regards,
Yi Liu
> >
> > struct vfio_iommu_type1_dma_map {
> > __u32 argsz;
> > __u32 flags;
> > #define VFIO_DMA_MAP_FLAG_READ (1 << 0) /* readable from device */
> > #define VFIO_DMA_MAP_FLAG_WRITE (1 << 1) /* writable from device */
> > __u64 vaddr; /* Process virtual address */
> > __u64 iova; /* IO virtual address */
> > __u64 size; /* Size of mapping (bytes) */
> > };
> >
> >>> + __u32 flags;
> >>> + __u16 addr_width;
> >>> + __u16 pasid_bits;
> >>> + __u32 padding;
> >>> + __u8 data[];
> >>> +};
> >>> +
> >>> +/*
> >>> + * struct iommu_nesting_info_vtd - Intel VT-d specific nesting info
> >>> + *
> >> spurious line
> >
> > yes, will remove this line.
> >
> > Regards,
> > Yi Liu
> >
> >>> + *
> >>> + * @flags: VT-d specific flags. Currently reserved for future
> >>> + * extension.
> >>> + * @cap_reg: Describe basic capabilities as defined in VT-d capability
> >>> + * register.
> >>> + * @ecap_reg: Describe the extended capabilities as defined in VT-d
> >>> + * extended capability register.
> >>> + */
> >>> +struct iommu_nesting_info_vtd {
> >>> + __u32 flags;
> >>> + __u32 padding;
> >>> + __u64 cap_reg;
> >>> + __u64 ecap_reg;
> >>> +};
> >>> +
> >>> #endif /* _UAPI_IOMMU_H */
> >>>
> >> Thanks
> >>
> >> Eric
> >
>
> Thanks
>
> Eric